چكيده به لاتين
Abstract:
Celiac disease is a self-immunity and chronic disorder of the small intestine caused by allergic reaction to gluten. This disease results in impaired nutrient absorption. As a result, it is considered as a malabsorption syndrom, and associated with severe consequences if not treated. Therefore, the diagnosis of celiac disease and determination of its grade for adopting an appropriate treatment is of the interest of pathologists and gastroenterologists. To this end, this thesis is conducted to determine the celiac grade by using a fuzzy cognitive map (FCM) based technique. This method improves the precision of classification by incorporating the grey system theory into the modeling process. Therefore, FCM and fuzzy grey cognitive map (FGCM) models are investigated as extension of FCM.
A limitation of FCMs is that they are unable to model the hesitancy introduced into a complex system due to imperfect facts, missing information, and indecision. To cope with this issue, FGCM model is presented. FGCM model deals with multi-concept environments. This model are based on the grey systems theory (GST), which has become a very effective mathematical theory for solving problems in a highly uncertain environment and with a discrete, small, and incomplete dataset. Due to the high uncertainty in the medical data, this extension of FCM that is resistant against uncertainty is used for celiac disease grading. By relying on the knowledge of skilled specialists and pathologists, the key features of celiac disease are extracted as the main concepts, and the models are designed. The results obtained by applying the proposed models on the real dataset with different grades of celiac disease verify the ability and effectiveness of these models. Because of better performance FGCM, active Hebbian learning (AHL) is proposed for FGCM learning to increase the accuracy and improve modeling performance. The simulation results show accuracies of 64.04%, 86.52% and 91.01% for the FCM, FGCM and trained FGCM. Among the different presented models, trained FGCM is displayed better abilities and uncertainty of the value of the output concept from this model is obtained lower than other models.
Keywords:
Fuzzy cognitive map; Fuzzy grey cognitive map; Active Hebbian learning algorithm; Celiac disease
